-2

I need to validate the correctness of heating/cooling cycle based on reading of temperature sensor in time.

Correct time series has a certain shape (number of ups and downs), lasts more or less the same amount of time, and has a certain max temperature which needs to be met during cycle. enter image description here

Typically the process is faulty when it is compressed or extruded in time. Has too low temperatures at peaks or in general the heating/cooling envelope is messed up. On the above picture I posted a simplified example of proper and wrong cycles of the process.

What classifier you would recommend for supervised learning models? Is unsupervised model at all possible to be developed for such scenario?

I am currently using calculation of max value of Temperature and Cross Correlation of 1 master typical proper cycle vs the tested one, but I wonder if there is a better more generic way to tackle the problem.

MatnikR
  • 181
  • 2
  • 16
  • do you want to have this running continuously or do you have logs that you want to validate? imho machine learning seems to be severe overengineering for this problem – Eumel Jul 30 '21 at 10:49
  • have you seen these related SO Qs, [1](https://stackoverflow.com/questions/11752727/pattern-recognition-in-time-series), [2](https://stackoverflow.com/questions/46784184/pattern-recognition-in-time-series-using-lstmpython), [3](https://stackoverflow.com/questions/64405608/what-is-the-correct-lstm-input-shape-for-a-multivariate-time-series)? – mnm Jul 30 '21 at 10:49
  • I am capturing the cycle data and start parallel verification process ending in email sent (faulty, proper plus plot). So it is defacto log based but with real time invitation after each cycle – MatnikR Jul 30 '21 at 10:51
  • @mnm - those questions tackle identifying a pattern - I am passed that I am able to identify the cycle as I have a separated sensor indicating start and stop of a process. I need to know if the shape of the pattern matches the template pattern or there are unacceptable deviations defined by example faulty cycles identified by experts – MatnikR Jul 30 '21 at 10:58

1 Answers1

2

imho machine learning is overengineering this problem, some banding and counting peaks seems to be the much easier approach to me.

Nontheless if you want machinelearning i would go with autoencoders for anomaly detection, examples can be found here or here.

Tl/dr: The idea is an autoencoder reconstructs the input though a very small bottleneck (i.e one value, could be the phase) so any current point will construct a good looking curve. Then it gets compared to the actual curve. If it fits all is good, if it doesnt you know something is not right.

Eumel
  • 1,298
  • 1
  • 9
  • 19